The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements structural protections against unauthorized command execution. The moonrakerGet function in handler.ts hardcodes the HTTP GET method and validates all requested paths against a strict regex ALLOWED_PATHS allowlist. This prevents the agent from issuing state-changing commands (like starting prints or heating) or sending arbitrary G-code to the printer.
[EXTERNAL_DOWNLOADS]: The skill communicates with a 3D printer at a local network IP address (default 192.168.1.139) or a URL specified in the MOONRAKER_URL environment variable. This is consistent with its intended function of monitoring hardware on a local network.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from an external printer API (Moonraker).
Ingestion points: handler.ts retrieves data from endpoints such as /printer/objects/query, /server/info, /server/history/list, and /server/files/list.
Boundary markers: No explicit boundary markers or "ignore" instructions are used when interpolating printer data into the human-readable summary.
Capability inventory: The skill is restricted to read-only API calls and writing JPEG image files to the workspace with enforced .jpg or .jpeg extensions.
Sanitization: The skill relies on the structured JSON format of the printer API and enforces file extension checks for snapshots, but does not perform content sanitization of the strings returned by the printer (e.g., filenames or status messages).

printer-monitor